Control of endothelial cell polarity and sprouting angiogenesis by non-centrosomal microtubules
Microtubules control different aspects of cell polarization. In cells with a radial microtubule system, a pivotal role in setting up asymmetry is attributed to the relative positioning of the centrosome and the nucleus. Here, we show that centrosome loss had no effect on the ability of endothelial cells to polarize and move in 2D and 3D environments. In contrast, non-centrosomal microtubules stabilized by the microtubule minus-end-binding protein CAMSAP2 were required for directional migration on 2D substrates and for the establishment of polarized cell morphology in soft 3D matrices. CAMSAP2 was also important for persistent endothelial cell sprouting during in vivo zebrafish vessel development. In the absence of CAMSAP2, cell polarization in 3D could be partly rescued by centrosome depletion, indicating that in these conditions the centrosome inhibited cell polarity. We propose that CAMSAP2-protected non-centrosomal microtubules are needed for establishing cell asymmetry by enabling microtubule enrichment in a single-cell protrusion.https://doi.org/10.7554/eLife.33864.001
Networks of blood vessels grow like trees. Sprouts appear on existing vessels, stretching out to form new branches in a process called angiogenesis. The cells responsible are the same cells that line the finished vessels. These “endothelial cells” start the process by reorganizing themselves to face the direction of the new sprout, changing shape to become asymmetrical, and then they begin to migrate.
Beneath the surface, a network of protein scaffolding supports each migrating cell. The scaffolding includes tube-like fibers called microtubules that extend towards the cell membrane and organize the inside of the cell. Destroying microtubules damages blood vessel formation, but their exact role remains unclear.
A structure called the centrosome can organize microtubules within cells. The centrosome was generally believed to act like a compass, pointing in the direction that the cell will move. Microtubules can anchor to the centrosome, and this structure is thought to play an important role in cell migration. Yet, many microtubules organize without it; these microtubules instead are organized by a compartment of the cell called the Golgi apparatus and are stabilized by a protein named CAMSAP2.
Martin et al. now report that removing the cells’ centrosomes did not affect cell migration, but getting rid of CAMSAP2 did. Analysis of cell shape and movement in cells grown in the laboratory and in living animals revealed that cells cannot become asymmetrical, or “polarize”, and migrate without CAMSAP2.
In a two-dimensional wound-healing assay, a sheet of cells originally grown from the vessels of a human umbilical cord was scratched, and a microscope was then used to record the cell’s movement as they repaired the injury. Normally, the cells on either side move in a straight line using their microtubules, and though the process was not affected in cells without centrosomes, it was in those without CAMSAP2.
Even more striking results were seen in three-dimensional assays. When the same blood vessel cells from human umbilical cords are grown as spheres inside collagen gels, they form sprouts as they would in the body. Without CAMSAP2, the cells could not organize their microtubules and they were unable to elongate in one direction and form stable sprouts. Lastly, depleting CAMSAP2 also prevented the normal formation of blood vessels in zebrafish embryos.
Taken together, these findings change our understanding of how microtubules affect cell movement and how important the centrosome is for this process. Further work could have an impact on human health, not least in cancer research. Tumors need a good blood supply to grow, so understanding how to block blood vessel formation could lead to new treatments. Microtubules are already a target for cancer therapy, so future work could help to optimize the use of existing drugs.https://doi.org/10.7554/eLife.33864.002
Cell polarization is a prerequisite for virtually every specialized cellular process. By analogy with the compass on a ship navigating to its destination, the centrosome has been assumed to play a central role in governing cell polarity (Bornens, 2012), including two-dimensional (2D) mesenchymal migration, in which cells are organized into an extending leading edge and a contractile cell rear (Ridley et al., 2003; Tang and Marshall, 2012). The widely accepted dogma connecting the centrosome and the direction of cell movement originated from the observation that during migration, the centrosome is typically positioned at the front of the nucleus (Gotlieb et al., 1981; Koonce et al., 1984; Kupfer et al., 1982; Malech et al., 1977). Such centrosome positioning is observed in different systems (Elric and Etienne-Manneville, 2014; Luxton and Gundersen, 2011), although it does not apply to certain specific cell types (Luxton and Gundersen, 2011; Yvon et al., 2002) or in the presence of environmental constraints (Doyle et al., 2009; Pouthas et al., 2008; Schütze et al., 1991). Despite the scarcity of direct experimental evidence, currently based on laser ablation of the pericentrosomal area (Koonce et al., 1984; Wakida et al., 2010) or indirect interference with centrosome localization (Dujardin et al., 2003; Etienne-Manneville and Hall, 2001; Levy and Holzbaur, 2008; Schmoranzer et al., 2009), the orientation of the nucleo-centrosomal axis is commonly regarded as a major cell polarity determinant.
Microtubules (MTs) are thought to support cell polarity by forming an asymmetrical network (Etienne-Manneville, 2013). This asymmetry can potentially be generated by specific positioning of the centrosomal anchor of a radial MT network relative to other cell structures, such as the nucleus, but also by MTs that do not originate from the centrosome (Akhmanova and Hoogenraad, 2015; Alieva et al., 2015; Vinogradova et al., 2009). The minus ends of centrosome-independent MTs can associate with the members of the calmodulin-regulated spectrin-associated protein (CAMSAP)/Patronin/Nezha family (Akhmanova and Hoogenraad, 2015). In mammalian cells, CAMSAP2, together with CAMSAP3 in certain cell types, binds to free, non-centrosomal MT minus ends and promotes their stabilization (Jiang et al., 2014; Tanaka et al., 2012). Interestingly, CAMSAP2 has recently been shown to participate in MT stabilization at the Golgi apparatus (Wu et al., 2016), a site that can function as an alternative MT-organizing centre (MTOC) (Zhu and Kaverina, 2013). In mammalian cells, the Golgi is often positioned close to the centrosome (Rios, 2014), and thus locates in front of the nucleus during cell migration (Koonce et al., 1984; Kupfer et al., 1982; Malech et al., 1977). As the central organizer of the secretory pathway, an anteriorly positioned Golgi is thought to support polarized transport needed to sustain directional migration (Yadav and Linstedt, 2011). Consistently, Golgi disorganization without MT disassembly (Bisel et al., 2008; Hurtado et al., 2011; Yadav et al., 2009), as well as the loss of Golgi-associated MTs (Hurtado et al., 2011; Miller et al., 2009) prevent proper polarized Golgi trafficking without affecting global secretory properties but lead to defects in directional cell movement. Notably, investigation of the contributions of centrosomal and Golgi-originating MT populations to Golgi organization indicated that the role of the centrosome was restricted to facilitating assembly of an integral Golgi apparatus (Vinogradova et al., 2012).
During angiogenesis, the process of new blood vessel development, endothelial cells (ECs) respond to external cues by coordinating numerous activities, including proliferation, sprouting, migration, lumen formation and anastomosis (Geudens and Gerhardt, 2011; Potente et al., 2011). The first step of vessel formation, outward cell sprouting, requires that the tip cell, which will guide growth of a new vessel, breaks symmetry by extending protrusions toward guidance cues (Geudens and Gerhardt, 2011; Lee and Bautch, 2011). MT growth dynamics has been shown to be required for the formation and maintenance of angiogenic structures (Bayless and Johnson, 2011; Myers et al., 2011), whereas its regional regulation has been implicated in directional EC migration (Braun et al., 2014). Vascular sprouting and repolarization are also affected by supernumerary centrosomes, a hallmark of tumor ECs that impacts on MT nucleation and dynamics (Kushner et al., 2014; Kushner et al., 2016).
The ability of ECs to polarize can be explored in 2D, where cells respond to a monolayer wound by developing a mesenchymal motile phenotype typified by a front-rear asymmetry and anterior centrosome positioning (Gotlieb et al., 1981). When cultured in 3D collagen gels, ECs form branched tubular structures (Koh et al., 2008), and when spheroids of ECs are embedded in a 3D matrix, they develop sprouts that closely reproduce the first step of formation of capillaries from pre-existing vessels (Pfisterer and Korff, 2016).
Here, we used a combination of 2D and 3D endothelial models to investigate the role of MT organization in cell polarization. Challenging the prevailing view, our data showed that the centrosomal MT population is dispensable and insufficient for EC migration and sprouting. In contrast, silencing of CAMSAP2, which results in disappearance of non-centrosomal MTs, profoundly perturbed the ability of ECs to form vascular sprouts in vitro as well as in vivo, in zebrafish embryos. Detailed analysis showed that non-centrosomal MTs are required to allow MT redistribution in a single-cell protrusion and thus enable polarized trafficking, directional stabilization of protrusions and persistent migration.
The centrosome is not essential for endothelial migration and sprouting
Using 2D monolayer wound healing assay, we confirmed the anterior position of the centrosome and the Golgi apparatus in migrating ECs and also extended this observation to tip ECs sprouting from a spheroid in 3D (Figure 1—figure supplement 1A). To assess the impact of centrosome removal on these processes, we took advantage of the Plk4 inhibitor centrinone, which prevents centriole duplication and leads to centrosome depletion (Wong et al., 2015). Efficient centrosome elimination was confirmed by staining with different markers of centrioles or pericentriolar material (PCM) (Figure 1—figure supplement 1B). Loss of centrosomes as focal points of MT organization was also visible by super-resolution imaging of the MT networks in fixed cells and by tracing growing EB3-GFP-positive MTs in live cells (Figure 1A,B). Importantly, the centrosomal arrangement of MTs in control cells was more obvious in EB3-GFP tracings than in MT images (Figure 1A,B), suggesting that the centrosome plays a more important role in nucleating new MTs than in anchoring their minus ends, and that even in control cells, a significant MT population is non-centrosomal. The perinuclear MT density was partly associated with the Golgi apparatus, which was mildly enlarged and less compact in centrinone-treated cells (Figure 1—figure supplement 1C). Nocodazole washout assays showed that the centrosome was the major nucleation site during MT reassembly in control cells, while this function was taken over by the Golgi membranes in centrinone-treated cells (Figure 1—figure supplement 1D). In spite of these differences, MT density, the density of growing, EB3-positive MT plus ends, parameters of MT plus end growth and the levels of different tubulin post-translational modifications were not affected (Figure 1A,B,C, Figure 1—figure supplement 2A,B), and in contrast to a centrosome excess (Godinho et al., 2014), the distribution and intensity of cell-cell adhesion markers was unchanged (Figure 1—figure supplement 2D ) (). The abundance of MT minus-end stabilizing protein CAMSAP2 was mildly but not significantly increased, and the area occupied by CAMSAP2 stretches was enlarged (Figure 1—figure supplement 2A,C). This is consistent with the enlarged Golgi, to which many CAMSAP2 stretches attach (Wu et al., 2016), and possibly with an increased stabilization of non-centrosomal MTs that compensate for the loss of the centrosomal ones.
Figure 1—source data 1
2D migration was not perturbed in centrinone-treated ECs (Figure 1D, Figure 1—figure supplement 2E). ECs in the migrating front still showed proper polarized organization with their Golgi positioned toward the direction of migration (Figure 1D). In a more physiological 3D context, centrosome removal did not affect the emergence of EC sprouts out of spheroids grown in collagen matrix (Figure 1E). The few remaining centrosome-containing ECs were not enriched in the sprouts, and the Golgi apparatus was properly polarized in sprouting tip ECs devoid of centrosomes (Figure 1F). To confirm these results, we depleted centrosomes by knocking down CPAP, a factor essential for centriole duplication (Kohlmaier et al., 2009; Schmidt et al., 2009; Tang et al., 2009). Also using this approach, we could efficiently remove centrosomes in the majority of cells without affecting EC polarization in 3D and sprouting from spheroids (Figure 1G,H, Figure 1—figure supplement 2F,G). These results indicate that centrosome is dispensable for endothelial polarization and movement.
CAMSAP2 is required for maintaining non-centrosomal MTs and cell migration in ECs
If the centrosomal MTs are dispensable for cell motility in ECs, their function must be taken over by non-centrosomal MTs, which in mesenchymal cells are stabilized by CAMSAP2 (Jiang et al., 2014). Interestingly, CAMSAP2 expression was transiently enhanced upon treatment with the angiogenic factor VEGF, and the length and number of CAMSAP2-decorated MT stretches was increased (Figure 2A,B, Figure 2—figure supplement 1A), suggesting that non-centrosomal MTs might play a role in angiogenesis. To test this idea, we silenced CAMSAP2 (Figure 2—figure supplement 1B), thereby generating a mostly centrosome-anchored, radial MT array (Figure 2C,D, Figure 2—figure supplement 1D). This treatment had no effect on the MT density or parameters of MT plus end dynamics (Figure 2D, Figure 2—figure supplement 1C). The abundance of post-translationally modified forms of tubulin was also unchanged (Figure 2—figure supplement 1E), showing that ECs are in this respect different from U2OS cells, where loss of detyrosinated MTs was seen upon CAMSAP2 depletion (Jiang et al., 2014). MT nucleation from the centrosome was unaffected, in line with fact that CAMSAP2 shows no colocalization with the centrosome (Figure 2—figure supplement 1F,G). Since MT properties were largely unaltered, the motility of the motor protein kinesin-3 KIF13B, which is known to play an important role in transporting VEGF receptor in ECs (Yamada et al., 2014), was also unchanged (Figure 2—figure supplement 1H).
Figure 2—source data 1
CAMSAP2 depletion caused a moderate but significant decrease in cell migration in 2D (Figure 2E, Figure 2—figure supplement 1I). Strikingly, the defect was much more pronounced during 3D sprouting: ECs depleted of CAMSAP2 were able to form only shorts sprouts and form long protruding structures (Figure 2F,G). This strong impairment was not related to cell viability or cell cycle progression defects (Figure 2—figure supplement 2A), and CAMSAP2 depletion in non-cycling ECs still severely affected formation of long sprouts (Figure 2—figure supplement 2B). Rescue experiments using expression of a siRNA-insensitive CAMSAP2 construct confirmed the specificity of the phenotype (Figure 2H), with ECs positive for the rescue construct populating growing sprouts in a CAMSAP2 knockdown background (Figure 2I). We also analyzed the potential involvement of CAMSAP3 and found that it was only weakly expressed in ECs and its depletion did not aggravate the phenotype of CAMSAP2 knockdown (Figure 2—figure supplement 2C–E). These data point to an important and specific role of CAMSAP2 in EC morphology in 3D.
CAMSAP2 is required for stabilization of one major cell protrusion
To understand the poor ability of CAMSAP2-depleted cells to form long sprouts in 3D, we set out to characterize this process in more detail. We found that early stages of sprout formation were not affected much by CAMSAP2 depletion; however, the differences gradually increased over time, as, in contrast to the control situation, spheroids silenced for CAMSAP2 were unable to increase the number and especially the length of the sprouts (Figure 3—figure supplement 1A). In the absence of CAMSAP2, endothelial spheroids are thus capable of initiating protrusions but are unable to mature them into larger and more stable, MT-populated structures (Figure 3A).
Figure 3—source data 1
In line with this idea, when individual isolated ECs were cultured in a collagen matrix, where they extended protrusions in different directions and fused into a tubulogenic network, CAMSAP2 depletion did not prevent the establishment of a vascular plexus (Figure 3—figure supplement 1B). Protrusive activity of isolated ECs measured by their elongation (inverse of circularity) and the total and average protrusion length was not affected by CAMSAP2 knockdown (Figure 3B, Figure 3—figure supplement 1C). However, such CAMSAP2-depleted ECs bearing a centrosome-centered MT array (Figure 3—figure supplement 1D) had a different organization of protrusions. Whereas control ECs had a restricted number of protrusions, with a single predominant one, CAMSAP2-depleted ECs displayed multiple short protrusions (Figure 3B,C). In contrast to control ECs that had most of their protrusions, and especially the longest ones, aligned in one direction, CAMSAP2-depleted cells displayed protrusions that were more radially dispersed, irrespective of their length (Figure 3D, Figure 3—figure supplement 1E,F).
To understand the origin of this phenotype, we performed live recording of protrusion formation. In contrast to control ECs, which, after having formed several small transient protrusions, stabilized and extended one or two of them, CAMSAP2-depleted ECs were unable to accomplish this transition and to elongate in a single direction (Figure 3E). Quantification of protrusion dynamics using a radial reslice representation (Figure 3F; every black signal represents the presence of a protrusion at a given time point at a given radial position) confirmed these observations and pointed to dramatically lower protrusion persistence after CAMSAP2 inactivation (Figure 3F), while the total protrusion number was not affected by CAMSAP2 depletion (Figure 3—figure supplement 1G).
We hypothesized that in CAMSAP2-depleted cells with a radial MT array, MTs cannot become enriched in one protrusion, and found that this indeed was the case (Figure 3G), strongly suggesting that non-centrosomal MTs stabilize polarized elongated cell morphology by enabling MT enrichment in a single protrusion.
CAMSAP2 depletion phenotypes cannot be explained by changes in the actin cytoskeleton and cell contractility
The reduced protrusion persistence prompted us to examine the organization of acto-myosin cytoskeleton after CAMSAP2 silencing. In 2D-cultured CAMSAP2-depleted cells, we observed a modest increase in the density of F-actin cytoskeleton (Figure 4A), due to the presence of more stress fibers (Figure 4—figure supplement 1A). Beside this small difference, cells established normal polarized front-rear morphologies, as revealed by the presence of actin-enriched lamellipodia and myosin IIb-positive retracting cell edges, and the distribution patterns of active Rho and Rac1, as well as cell adhesion markers were normal (Figure 4A,B, Figure 4—figure supplement 1A,B,C). Importantly, actin cytoskeleton was unchanged in 3D environment in the absence of CAMSAP2, displaying intense peripheral cortical accumulation as in control situation (Figure 4C, Figure 4—figure supplement 1D).
Figure 4—source data 1
Myosin II-dependent cell contractility has been shown to inhibit protrusion formation in ECs (Fischer et al., 2009). In agreement with these observations, inhibition of myosin II either directly, with blebbistatin, or indirectly, with the inhibitor of the kinase ROCK (Y-27632), led to longer 3D protrusions in each condition (Figure 4—figure supplement 2A–D). However, in the absence of CAMSAP2, these treatments did not rescue the polarized elongated morphology typical of control ECs (Figure 4D, Figure 4—figure supplement 2A,C,E). Moreover, although decreased contractility facilitated protrusion persistence, CAMSAP2 depletion still severely reduced protrusion stability in the presence of the ROCK inhibitor (Figure 4E). Similarly, treatment of spheroids with the ROCK inhibitor or blebbistatin induced more and longer sprouts but failed to rescue sprouting impairment in CAMSAP2-depleted cells (Figure 4F, Figure 4—figure supplement 2F). Altogether, these data demonstrate that changes in acto-myosin cytoskeleton or contractility cannot explain the defects associated with CAMSAP2 silencing.
CAMSAP2 participates in Golgi polarization during 2D migration and 3D sprouting
While displaying normal front-rear morphologies, CAMSAP2-depleted cells were often unable to orient their lamellae in the direction of migration, suggesting polarity defects (Figure 4B, asterisks). Accordingly, CAMSAP2 inactivation resulted in a substantial drop in the directionality of cell movement (Figure 5A), which was associated with Golgi and centrosome mispositioning respective to the wound (Figure 5B, Figure 5—figure supplement 1A). This explained the global impairment of cell migration in spite of only a minor decrease in movement velocity (Figure 2—figure supplement 1I). Furthermore, the correlation between the positions of the Golgi and the centrosome, which always colocalized (Figure 5—figure supplement 1A), and the lamellipodia was strongly diminished after CAMSAP2 silencing, with some ECs having their leading edge in front of the Golgi and some not (Figure 5B). Forward-targeted post-Golgi vesicle trafficking, which is often considered as a key regulator of cell polarity during migration, was perturbed in the absence of CAMSAP2. Whereas in control ECs the majority of exocytotic vesicles labeled with the small GTPase Rab6 moved towards the leading edge, Rab6 trajectories were distributed more symmetrically in CAMSAP2-depleted ECs, while the other trafficking parameters were not affected (Figure 5C, Figure 5—figure supplement 1B). This likely reflected a more radial, symmetric MT array in these cells (Figure 2C, Figure 2—figure supplement 1D). Together, our 2D results argue in favor of a model where Golgi positioning does not directly dictate the orientation of the extending lamellae but is needed to stabilize directional persistence by controlling polarized trafficking.
Figure 5—source data 1
Forward positioning of the Golgi, which strongly co-localized with CAMSAP2 stretches, was even more striking in 3D (Figure 5—figure supplement 1C). In this 3D setting, CAMSAP2 proved to be essential for Golgi polarization in the direction of sprouting (Figure 5D). In line with our 2D observations, while control cells displayed highly asymmetric trafficking of MT-dependent Rab6 positive exocytotic vesicles, little asymmetry in trafficking was observed between protrusions of CAMSAP2-depleted cells (Figure 5E). Consistent with the inability of centrosome-anchored MTs to redistribute to a single protrusion (Figure 3G), the failure in polarization of the secretory trafficking, which is likely necessary to generate a stable leading cell edge (Schmoranzer et al., 2003; Stehbens et al., 2014; Yadav et al., 2009) could explain the inability of CAMSAP2 depleted ECs to form mature sprouts in 3D.
Loss of non-centrosomal MTs has a more severe impact than their detachment from the Golgi in 3D but not in 2D
The above results suggest that by protecting Golgi-tethered MTs, CAMSAP2 regulates proper Golgi positioning important for EC polarization. To test this idea, we made use of our recent findings showing that two proteins, AKAP450 and myomegalin (MMG), are needed for anchoring CAMSAP2-decorated MT minus-ends to Golgi membranes (Wu et al., 2016). As expected, depletion of AKAP450 or MMG (Figure 6—figure supplement 1A) displaced CAMSAP2 stretches from the Golgi (Figure 6A, Figure 6—figure supplement 1B). This redistribution had remarkably similar consequences for 2D migration compared to CAMSAP2 silencing: ECs that had lost their Golgi-attached CAMSAP2 were unable to maintain directionality and to orient their Golgi, resulting in migration deficiency (Figure 6B, Figure 6—figure supplement 1C,D). Interestingly, the situation differed when ECs were placed in a 3D context: whereas displacing non-centrosomal CAMSAP2-bound MT ends from Golgi by depleting AKAP450 or MMG had a negative impact on endothelial sprouting abilities, its effect was significantly milder than that of CAMSAP2 depletion, when most MTs were attached to the centrosome (Figure 6C). This suggests that in 3D sprouting, non-centrosomal MTs might have a role independent of their direct Golgi association.
Figure 6—source data 1
Centrosome removal promotes cell polarization in the absence of CAMSAP2
We reasoned that non-centrosomal MTs, which are not anchored to a single point, might redistribute more easily to create asymmetry, and therefore, centrosome removal in CAMSAP2-depleted ECs might improve their polarization potential by restoring a pool of non-centrosomal MTs. To test this idea, we silenced AKAP450, MMG and CAMSAP2 in ECs in combination with centrinone-induced centrosome depletion. Such ECs were viable and efficiently lost their centrosome, displaying a characteristic enlarged shape filled with a dense non-centrosomal MT array, even in the absence of CAMSAP2 (Figure 7A). The density of MTs and growing, EB3-positive MT ends was similar in all conditions (Figure 7B, Figure 7—figure supplement 1A). As we described in our previous study (Wu et al., 2016), acentrosomal cells showed enhanced recruitment of γ-tubulin to the Golgi, an effect that was abolished by depleting AKAP450 (Figure 7C,D). In line with these results, we observed abundant MT nucleation from the Golgi membranes after nocodazole washout in control, MMG- and CAMSAP2-depleted acentrosomal cells, while in AKAP450-depleted acentrosomal cells, MTs were nucleated from the cytoplasmic sites distinct from the Golgi membranes (Figure 7—figure supplement 1B). Such distribution of MT nucleation sites correlated with the recruitment of the pericentriolar material (PCM) marker Pericentrin to the Golgi in control, MMG- and CAMSAP2-depleted centrinone-treated cells, and its dispersion in the cytoplasm in AKAP450-depleted cells (Figure 7—figure supplement 1C).
Figure 7—source data 1
In agreement with our previous work in RPE1 cells (Wu et al., 2016), MT density was detached from the Golgi region after AKAP450 and MMG knockdown, because these proteins constitute a part of an essential link between MTs and the Golgi membranes (Figure 7A,D). In CAMSAP2-depleted ECs, centrinone treatment caused some disorganization of the Golgi, and MTs were not concentrated in the Golgi area either (Figure 7A,D).
The absence of the centrosome caused no additional reduction in the sprouting ability of AKAP450 or MMG-depleted ECs organized in spheroids (Figure 7E), indicating that CAMSAP2-stabilized non-centrosomal MTs are sufficient to support formation of elongated sprouts from spheroids to some extent even when they are not attached to the Golgi. Strikingly, the removal of centrosome in the absence of CAMSAP2 significantly rescued the sprouting potential of ECs (Figure 7E). These data support the idea that non-centrosomal MTs contribute positively to EC sprouting, while the centrosome is not only dispensable, but can also play an inhibitory role when CAMSAP2 is absent.
Non-centrosomal MTs are required to create protrusion asymmetry
Among all the situations analyzed, CAMSAP2-depleted cells which had centrosomes were the only ones which had a symmetric, strongly radial MT system, while this property was lost when these cells were treated with centrinone (Figure 7A, Figure 8A). As described above, CAMSAP2-depleted cells with centrosomes had symmetric radial protrusions in 3D (Figure 3B–D), but, remarkably, centrosome depletion restored their ability to generate one long dominant protrusion (Figure 8B, Figure 8—figure supplement 1). MMG-depleted cells, in which non-centrosomal MTs are present but not anchored at the Golgi performed in these assays just as well as control cells irrespective of their centrosome content (Figure 8B), and we found that although CAMSAP2-decorated minus ends were not enriched at the Golgi anymore after MMG depletion, they were still concentrated in the major protrusion (Figure 8C). Therefore, Golgi attachment is not a requirement for concentrating non-centrosomal MTs in one cell protrusion in 3D. This observation likely explains why centrosome removal rescues sprouting in CAMSAP2-delpeted cells: non-centrosomal MTs can concentrate in one protrusion (Figure 8D) even though they are not linked to Golgi membranes. Taken together, our data demonstrate that the presence of non-centrosomal MTs is essential for generating cell asymmetry required for the emergence of long EC sprouts, while attachment of these MTs to the Golgi, and the likely more efficient secretion associated with such an arrangement is beneficial but not essential.
Figure 8—source data 1
CAMSAP2 plays a role in sprouting angiogenesis in vivo
Finally, we addressed the role of CAMSAP2 in vivo using zebrafish as a model. Two CAMSAP2-encoding gene orthologues are present in zebrafish, camsap2a (ENSDARG00000062173) and camsap2b (ENSDARG00000059965). To analyze the role of CAMSAP2 in zebrafish vascular development, we designed splice-blocking antisense morpholinos (Figure 9—figure supplement 1A) to generate CAMSAP2-silenced embryos in the endothelial reporter line Tg(fli1a:eGFP). Embryos inactivated for CAMSAP2a or CAMSAP2b were viable, had no obvious morphological defects and normal somite development. We next focused on CAMSAP2b, because among CAMSAP2 orthologs, it showed the highest expression in the zebrafish endothelium and was associated with more severe vascular defects. In zebrafish, two waves of dorsal sprouting angiogenesis take place successively during vascular development (Ellertsdóttir et al., 2010; Isogai et al., 2003). The first one occurs at around 22 hr post fertilization (hpf) from the dorsal aorta and forms arterial intersegmental vessels. Another one takes place between 32 and 48 hpf from the cardinal vein and gives rise to venous intersegmental vessels and to parachordal lymphangioblasts, precursors of lymphangiogenic vessels (Figure 9A). After CAMSAP2b inactivation (Figure 9—figure supplement 1B), arterial intersegmental vessel formation was hardly altered, but the secondary EC sprouting was perturbed, giving rise to abnormal tortuous venous intersegmental vessels (Figure 9—figure supplement 1C). In CAMSAP2b morphants, we observed fewer secondary sprouts emerging from the cardinal vein at 34 hpf (Figure 9—figure supplement 1D). In addition, the fraction of venous sprouts that had fused with the neighboring arterial intersegmental vessel at 41 hpf was strongly reduced (Figure 9—figure supplement 1D), suggesting a defect in directional migration. Venous intersegmental vessels form after cardinal vein-derived secondary sprouts connect to primary arterial intersegmental vessels. Because secondary sprouts had difficulties to fuse with primary intersegmental vessels, we also found that the proportion of venous intersegmental vessels at 48hpf was reduced in CAMSAP2b morphants: whereas control embryos displayed a typical 50–50 arterial/venous intersegmental vessel ratio, fewer intersegmental vessels were connected to the cardinal veinand scored as venous intersegmental vessels in CAMSAP2b morphant embryos (Figure 9A,B). In addition, the alternative outcome of venous sprouting, that is parachordal lymphangioblasts formation, was alsoimpaired in the absence of CAMSAP2b, with less or aberrant parachordal lymphangioblasts in the morphants (Figure 9A, Figure 9—figure supplement 1E, asterisks). Importantly, injection of mRNA coding for human CAMSAP2 restored normal phenotypes in the majority of the morphants (Figure 9B, Figure 9—figure supplement 1F).
Figure 9—source data 1
To confirm these results, we performed live imaging of venous sprouting in Tg(Fli1ep:Lifeact-EGFP), a zebrafish line with F-actin labeled in the endothelium (Phng et al., 2013). In control conditions, venous ECs migrated in a highly directed manner to either fuse to the neighboring arterial intersegmental vessel or start assembling horizontal parachordal lymphangioblasts (Supplemental Videos 1 and 2, Figure 9C). In contrast, in CAMSAP2b morphants venous sprouts were very unstable and showed less directional persistence (Supplemental Videos 3 and 4, Figure 9C), sometimes resulting in the atypical fusion of two distinct sprouts with the same arterial intersegmental vessel (Supplemental Video 5). These observations were quantitatively validated by tracking the length and orientation of venous sprouts over time. In contrast to their regular extension in control animals, venous sprouts depleted of CAMSAP2b displayed a more erratic and less efficient growth (Figure 9D). This aberrant behavior is also illustrated in Figure 9E, where CAMSAP2b silencing is shown to be associated with a lot of shortening episodes (negative change in length) and a more variable sprout orientation (higher change in angle), resulting in a significantly lower growth and directional persistence (Figure 9F). Supporting a role for CAMSAP2 in venous angiogenesis, formation of the caudal vein plexus, a honeycomb-like structure arising from active ventral migration of venous ECs from the cardinal vein in the caudal region, was also perturbed in the morphants (Figure 9—figure supplement 1G). Altogether, these observations suggest a defect in guided migration during venous sprouting in CAMSAP2b-silenced zebrafish embryos, in agreement with our in vitro findings and supporting the idea that CAMSAP2 is involved in directional angiogenic sprouting in vivo.
While the acto-myosin cytoskeleton is crucial for generating protrusions, adhesions and contractile forces during cell migration, an anisotropic MT network strongly contributes to the establishment and maintenance of cell polarity. One current model explaining the generation of an asymmetric MT array involves forward positioning of its center, the centrosome, assumed to represent the main MTOC, together with the local regulation of MT plus end stability (Etienne-Manneville, 2013; Vinogradova et al., 2009). Here, we showed that although these regulatory processes could contribute to the asymmetry of the system, a centrosomal radial MT network was both completely dispensable and insufficient for the establishment of polarized cell morphology in soft 3D matrices (Figure 9G).
Previous work showed that the MT minus-end binding protein CAMSAP2 is a key player in the regulation of non-centrosomal MT minus ends in mammalian cells (Akhmanova and Hoogenraad, 2015; Jiang et al., 2014; Tanaka et al., 2012; Yau et al., 2014). Here, we uncovered the crucial role of CAMSAP2 in regulating cell polarity during endothelial sprout formation in 3D and persistent directional migration in 2D (Figure 9G). We think that this demonstrates the important role of non-centrosomal MTs in these processes, because, although we cannot exclude that this protein has alternative functions, for example in controlling motor-based transport or signaling, we did not find any direct evidence supporting such functions in our experiments. Furthermore, although MTs are known to regulate actin organization, and MT destabilization promotes myosin II-dependent contractility, which affects protrusion formation (Etienne-Manneville, 2013), the protrusions formed in the absence of CAMSAP2 exhibited unchanged F-actin network. This was in line with our observation that the loss of CAMSAP2 had no impact on the overall microtubule density, plus end dynamics or modifications. Our observation that pharmacological inhibition of contractility failed to rescue the persistency and polarized organization of cell protrusions after CAMSAP2 depletion further supports the view that the observed cell morphology and migration defects were not caused by changes in the actin cytoskeleton. Importantly, CAMSAP2-depleted ECs could still form lamellipodia and retracting cell rear on 2D surfaces and initiated protrusions when cultured in 3D. Whether this reflects a sufficient degree of asymmetry supported by the centrosomal MT network, or a lack of MT involvement in these actin-based processes, as it has been suggested in 2D models (Siegrist and Doe, 2007), deserves additional investigation. However, both in 2D and 3D, loss of non-centrosomal MTs interfered with persistence of lamellipodia and elongation of cell protrusions, explaining migration defects.
CAMSAP2 played a much more prominent role in 3D than in 2D environment and even displayed different phenotypes in the two distinct 3D settings we used: the absence of CAMSAP2 severely reduced sprouting from spheroids, while the protrusive activity of single isolated ECs in collagen matrix was unaffected. It is likely that the distinct degree of polarity required in these assays could explain this difference. The inability of spheroids to maintain and extend long sprouting structures will culminate in their collapse (as seen in Figure 3—figure supplement 1A), whereas isolated ECs could form protrusions in any direction. However, also in the latter case, the radial MT network in CAMSAP2-silenced ECs could not support polarized elongated morphology characteristic for control cells. In our recent study (Bouchet et al., 2016), we showed that in mesenchymal cells, the initial formation of protrusions is MT-independent, but the extension and stabilization of long protrusions occurs only if they are filled with MTs. The data presented in the current study indicate that in order to acquire an elongated polarized morphology, 3D-cultured ECs need to be able to concentrate their MTs in one protrusion. If the intrinsic centrosome-driven symmetry of the MT network dominates, it inhibits this process, forcing the cells to acquire a non-polarized ‘starfish’-like shape, which is incompatible with efficient cell elongation and translocation in 3D matrix, as observed in the spheroid sprouting assay. While our results focus on one the very first steps of sprouting angiogenesis, that is the extension and stabilization of a protrusion that is required for effective outward migration, other important angiogenic behaviors involving polarity, as multicellular sprout growth and vessel lumenization were not addressed in this study and deserve further investigation.
Non-centrosomal MTs can form intrinsically asymmetric networks through their attachment to the Golgi complex (Vinogradova et al., 2009; Zhu and Kaverina, 2013). Indeed, confirming previous results (Roubin et al., 2013; Zhu and Kaverina, 2013), we showed that the depletion of Golgi-originating MTs dependent on AKAP450 and MMG fully recapitulated CAMSAP2 inactivation-related defects in 2D assays. This suggests that the Golgi-anchored MT population, required for the proper Golgi positioning and polarized trafficking, is an important determinant of directional 2D migration. However, in the 3D spheroid assays, the impact of AKAP450 or MMG knockdown was far less severe than that of CAMSAP2 depletion. In fact, an elongated morphology in isolated ECs in 3D could be established when MMG was depleted and Golgi-attached MTs were absent. This suggests that non-centrosomal MTs can function independently of the Golgi anchoring. This idea is supported by the observation that the phenotype caused by the loss of CAMSAP2 could be partially rescued by suppressing formation of a symmetric centrosome-associated MT aster and thus reverting to a non-centrosomal MT array, albeit one lacking Golgi-MT attachments. In fact, the remarkable similarity between the phenotypes associated with CAMSAP2 and MMG depletion after centrinone treatment, leading to similar MT organization, strengthens the idea that the presence of non-centrosomal MTs per se, rather than their Golgi attachment might be crucial to support at least some degree of 3D sprout formation. The redistribution of MT nucleating and anchoring PCM complexes to the Golgi and cytoplasm likely contributes to the generation of such arrays in centrinone-treated cells. It is possible that asymmetrical cortical stabilization of MTs, regulated by extrinsic signals, can be sufficient to polarize the MT network independently of Golgi anchoring, if this polarity is not perturbed by the presence of a potent symmetric MT-anchoring structure such as the centrosome.
We recently showed that a single Golgi apparatus can assemble in the absence of centrosomal and Golgi-derived MTs (Wu et al., 2016), suggesting that non-centrosomal MTs that are detached from the Golgi membranes can still regulate Golgi organization. It is therefore possible that non-centrosomal MTs that are not anchored at the Golgi can still control Golgi positioning and ensure polarized secretion, as the Golgi is typically oriented toward the main protrusion in these conditions (Figure 8C). In this situation, the Golgi apparatus and non-centrosomal MTs, although not permanently connected, likely exert a positive feedback on each other. The Golgi, which in the absence of the centrosome concentrates γ-tubulin and some other PCM components, serves as the major MT nucleation site, albeit a one which cannot tether the MTs it nucleates. MTs, in turn, serve as directional tracks for localizing Golgi membranes, and increasing MT density in one protrusion will help to maintain the Golgi at the base of this protrusion.
The mechanism of MT minus end stabilization in the absence of both the centrosome and CAMSAP2 is still an open question. An interesting candidate is ninein, a protein involved in MT minus-end organization in epithelial cells (Moss et al., 2007), whose worm homolog can act on the same pathway as CAMSAP homologs (Wang et al., 2015). Interestingly, ninein has been described to relocalize from the centrosome to the cytoplasm upon triggering angiogenesis and to participate in endothelial morphogenesis (Matsumoto et al., 2008).
In line with our in vitro results, CAMSAP2 inactivation in zebrafish perturbed the directional and persistent migration of ECs sprouting dorsally from the cardinal vein. Interestingly, loss of CAMSAP2 had no effect on EC sprouting form the dorsal aorta, thus suggesting that venous sprouting might be more sensitive to the lack of CAMSAP2. Recent evidence suggests that ECs from different vascular beds are differentially regulated and use different mechanisms (Franco et al., 2016; Rocha and Adams, 2009). It is possible that the secondary wave of sprouting from the cardinal vein is more dependent on intrinsic polarity mechanisms than the arterial sprouting, for instance, due to differential participation of supportive cells and therefore different need for cytoskeleton-based processes.
One of the most surprising findings of this study is the lack of importance of the centrosome in endothelial polarization. Although we do not exclude that centrosomal MTs are participating in polarity establishment in control cells, where the centrosome is in fact the major nucleating factor, we provide evidence that its MT anchoring activity is dispensable, not sufficient and should be counterbalanced by a non-centrosomal MT population. As it was already described for other organisms (Bornens, 2012), our results support the view that the centrosome has no crucial function in many types of animal tissues. Whereas the role of centrosome localization in determining neuronal polarity in vivo has been heavily debated (Kuijpers and Hoogenraad, 2011), an inhibitory role for a radial centrosomal MT organization has recently been suggested in epithelia (Noordstra et al., 2016). Altogether, our findings support the concept that polarity induction requires a switch to an asymmetric MT network, which might involve participation of centrosome-independent MT minus end stabilizing factors and centrosome inactivation.
Materials and methods
Antibodies, reagents and constructsRequest a detailed protocol
We used rabbit polyclonal antibodies against CAMSAP2 (Novus, Littleton, CO, NBP1-21402), CEP135, acetylated tubulin, polyglutamylated tubulin and γ-tubulin (Sigma-Aldrich, St Louis, MO, SAB4503685, T7451, T9822 and T3559), CDK5RAP2 (BethylLaboratories, Montgomery, TX, A300-554A), detyrosinated tubulin (Abcam, UK, ab48389), EB3 (Stepanova et al., 2003) and myomegalin isoform 8 (MMG8) (Wang et al., 2014), goat polyclonal antibodies against MYOSIN IIb and PCM1 (Santa-Cruz biotechnology, Dallas, TX, SC-47205 and SC-50164), mouse monoclonal antibodies against GM130, Pericentrin, EB1, VE-Cadherin, ZO-1, AKAP450 and KU80 (BD Biosciences, San Jose, CA, 610823, 611815, 610535, 610252, 610966, 611518 and 611360), CAMSAP3, α-tubulin and γ-tubulin (Sigma-Aldrich, SAB4200415, T5168 and T6557), NEDD1 (Abnova, Taiwan, H00121441-M05) and rat monoclonal antibodies against α-tubulin YL1/2 (Pierce, Waltham, MA, MA1-80017). Rabbit polyclonal antibody against CPAP was a kind gift of Dr. P.Gönczy (Swiss Institute for Experimental Cancer Research, EPFL, Lausanne, Switzerland).
For western blots, we used the following secondary antibodies: IRDye 800CW/680 LT Goat anti-rabbit and anti-mouse (Li-Cor Biosciences, Lincoln, LE). For immunofluorescence, Alexa Fluor 488,–594 and −647 conjugated goat antibodies against rabbit, rat and mouse IgG were used as secondary antibody (Molecular Probes, Eugene, OR), together with Alexa Fluor 488/594 phalloidin and NucRed Live 647 (Molecular Probes) and DAPI (Sigma-Aldrich).
For STED imaging, Atto 647N Phalloidin and Abberior STAR 635P anti-mouse antibodies (Sigma-Aldrich) were used.
High-concentration rat tail Collagen I was from Corning (Corning, NY), Phorbol 12-myristate 13-acetate (PMA), nocodazole, Y-27632 and DAPI were from Sigma-Aldrich and human recombinant Fibroblast Growth Factor (FGF) and Vascular Endothelial Growth Factor (VEGF) were from Peprotech (UK) and blebbistatin was from Enzo Life Sciences (Belgium). Centrinone was a kind gift of Dr. A.Shiau and Dr. T. Gahman (Small Molecule Discovery Program, Ludwig Institute for Cancer Research, San Diego).
The CAMSAP2 siRNA insensitive construct consists of a truncation of the first 232 amino acids of human CAMSAP2 generated by a PCR-based strategy and cloned into peGFP-C1 (Clonetech, Montain view, CA). The zebrafish rescue construct is described below in the zebrafish section.
The constructs coding for Rab6A and EB3 in peGFP-C2 were described elsewhere (Matanis et al., 2002) and Stepanova et al., 2003), GFP-KIF13B was a gift from Dr.A.Chishti (University of Illinois, Chicago).
The Rho biosensor coding plasmid pLenti-RhoA2G (Addgene plasmid # 40179) is a gift of Dr.O. Pertz, (University of Basel, Switzerland) and the pLVIN-Rac1-bs Rac1 biosensor plasmid was described elsewhere (Bouchet et al., 2016).
siRNAsRequest a detailed protocol
We used the following siRNAs purchased from Sigma siRNA CAMSAP2 #1, 5’- GAATACTTCTTGACGAGTT-3’ (Jiang et al., 2014) siRNA CAMSAP2 #2, 5’- GTACTGGATAAATAAGGTA-3’ (Jiang et al., 2014) siRNA CAMSAP3, 5’-GCATTCTGGAGGAAATTGA-3’ (Noordstra et al., 2016) siRNA AKAP450, 5’-AUAUGAACACAGCUUAUGA-3’ (Hurtado et al., 2011) siRNA MMG, 5’-AGAGCGAGATCATGACTTA-3’ (Roubin et al., 2013) siRNA CPAP, 5’- AGAAUUAGCUCGAAUAGAA-3’ (Tang et al., 2009) siRNA Luciferase control, 5’-CGTACGCGGAATACTTCGA-3’
Cell culture and treatmentRequest a detailed protocol
Human Umbilical Vein Endothelial Cells (HUVECs) were obtained from Lonza and grown in endothelial basal medium (EGM-2) supplemented with growth supplements (SingleQuots, Lonza, Switzerland): 2% Fetal Bovine Serum (FBS), human Epidermal Growth Factor (hEGF), Vascular Endothelial Growth Factor (VEGF), R3-Insulin-like Growth Factor-1 (R3-IGF-1), Ascorbic Acid, Hydrocortisone human Fibroblast Growth Factor-Beta (hFGF-β), Heparin, Gentamicin/Amphotericin-B (GA).
HUVECs authentication was guaranteed by Lonza through identity and quality control and testing, including against mycoplasma, bacteria, yeast, and fungi. Only low passage cells (between passages 3 and 7) were used.
Plasmids and siRNA were, respectively, nucleofected using Amaxa technologies with the HUVEC nucleofector kit (Lonza) and transfected with GeneTrans II (MoBiTec, Germany) reagents according to the manufacturers' protocols. CPAP depletion was achieved through four successive rounds of siRNA transfection every 3 days.
Amaxa was additionally used in the specific case of siRNA transfection of centrinone-treated HUVECs.
HUVECs were treated with 125 nM centrinone for 9 days, including the time needed for functional assay. During treatment, non-treated and centrinone-treated HUVECs were passaged every 2–3 days to keep a 60–90% confluency. All functional assays involving centrinone were done in the presence of 10 nM Thymidine.
For VEGF treatment, HUVECs were starved in 0.5% serum containing medium for 36 hr before addition of 50 ng/ml of VEGF.
Nocodazole-induced MT complete disassembly was performed by treating HUVECs with 10 µM nocodazole for 2 hr at 37˚C, followed by 1 hr at 4˚C. Washout (WO) was then carried out by two washes with cold and two washes with warm medium.
Y-27632 and blebbistatin were used at 10 µM and 50 µM, respectively.
Cell proliferation assaysRequest a detailed protocol
HUVECs were trypsinized 48 hr after transfection and counted by trypan blue staining for quantification of the doubling time. Alternatively, HUVECs were stained with phalloidin together with DAPI to identify mitotic cells and their percentage was calculated to determine the mitotic index.
Western blottingRequest a detailed protocol
HUVECs total extracts were prepared in RIPA buffer (10 mM Tris-HCl pH 8, 140 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 0,1% SDS, protease inhibitor cocktail (Complete - Sigma)).
SDS-PAGE and Western blot analysis were performed according to standard procedures and developed with the Odyssey technology (Li-Cor Biosciences). Densitometric analysis was done using the ‘gel analysis’ plug-in of ImageJ.
Monolayer wound healing assayRequest a detailed protocol
A confluent HUVEC monolayer was scratched using a sterile P200 tip to create a cell-free zone. Fields were photographed just after injury and 8 hr later. Quantification of cell migration was made by measuring the percentage of area recovery using ImageJ software. Alternatively, phase-contrast live imaging was performed.
3D individual seeding assayRequest a detailed protocol
Single HUVECs were seeded into 2.5 mg/ml collagen pH buffered gels overlaid with complete medium supplemented with 50 ng/ml FGF, VEGF and PMA. For global vascular network assessment, 2 × 106 cells/ml were embedded for 48 hr and wide-field fluorescence imaging was done on EVOS cell imaging system (ThermoFisher Scientific, Waltham, MA). For cell morphology analyses, 0.5 × 106 cells/ml embedding experiments were submitted to phase-contrast live imaging or processed for immunostaining after 24 hours.
3D spheroid sprouting assayRequest a detailed protocol
The spheroid sprouting assay was performed as previously described (Martin et al., 2013): HUVEC spheroids were generated overnight by culturing endothelial cells in complete medium containing 20% methylcellulose in non-adherent 96 well plates. Harvested spheroids were then embedded into 2 mg/ml collagen pH buffered gels overlaid with complete medium supplemented with 40 ng/ml FGF and 50 ng/ml PMA. Angiogenic activity was quantified by measuring the cumulative length of the sprouts that had grown out of each spheroid, their mean number and length, 24 hr after embedding using ImageJ software on bright field images. The sprouts that were originating from secondary branching and the ones that were not in focus in the pictures were omitted from the analysis.
Immunofluorescence stainingRequest a detailed protocol
For 2D (/3D) staining, HUVECs were fixed with −20°C methanol for 10 min or with 4% PFA for 12 (/20) minutes at RT, permeabilized with 0.15% Triton X-100 in phosphate buffered saline (PBS) (/PBS-glycine 0.1M) for 2 (/45) minutes, sequentially incubated 1 hr in blocking buffer 2% BSA, 0.05% Tween-20 in PBS (/2% BSA, 0.05% Tween-20, 0.2% Triton X-100, 0.05% NaN3 in PBS), 1 (/4) hr in primary antibody in blocking buffer, 1 (/1.5) hour in secondary antibody, Alexa-conjugated phalloidin and DAPI in blocking buffer. After several washes, slides (/dismounted 3D gel plugs) were air-dried and mounted in Vectashield mounting medium (Vector laboratories, Burlingame, CA).
For immunostaining of HUVECs submitted to wound-healing assay, samples were fixed after 6 hr of migration.
For STED imaging, HUVECs were pre-extracted 45 s in extraction buffer (PEM80, 0.3% Triton-X100, 0.15% gluteraldehyde) at 37˚C, fixed with 4% PFA for 12 min at 37˚C, permeabilized with 0.2% Triton X-100 for 10 min at RT and then submitted to the same 2D protocol as above except the removal of Tween-20 in the buffers.
Image acquisitionRequest a detailed protocol
Bright-field images were collected on an EVOS cell imaging system (ThermoFisher Scientific) and phase-contrast live cell imaging was performed on a Nikon Ti equipped with a perfect focus system Nikon), a super high pressure mercury lamp (C-SHG1, Nikon, Japan), a Plan Fluor DLL 10x NA 0.3 (Ph1), a CoolSNAP HQ2 CCD camera (Photometrics, Tucson, AZ), a motorized stage MS-2000-XYZ with Piezo Top Plate (ASI, Eugene, OR) and a stage top incubator (Tokai Hit, Japan) for 37°C/5% CO2 incubation. The microscope setup was controlled by Micro-manager software.
For fluorescence imaging of 2D fixed samples and 3D fixed and live samples, including zebrafishes, Z-series images were collected with spinning disk confocal microscopy on a Nikon Eclipse Ti microscope equipped with a perfect focus system (Nikon), a spinning disk-based confocal scanner unit (CSU-X1-A1, Yokogawa, Japan), an Evolve 512 EMCCD camera (Roper Scientific, Trenton, NJ) attached to a 2.0X intermediate lens (Edmund Optics, Barrington, NJ), a super high pressure mercury lamp (C-SHG1, Nikon), a Roper scientific custom-ordered illuminator (Nikon, MEY10021) including 405 nm (100 mW, Vortran), 491 nm (100 mW, Cobolt) 561 nm (100 mW, Cobolt) and 647 nm (100 mW, Cobolt) excitation lasers, a set of BFP, GFP, RFP and FarRed emission filters (Chroma, Bellows Falls, VT) and a motorized stage MS-2000-XYZ with Piezo Top Plate (ASI). The microscope setup was controlled by MetaMorph. Images were acquired using Plan Fluor 20x MI NA 0.75 and Plan Apo VC 60x NA 1.4 oil objectives and Apo LWD λS 40x water immersion objective. When necessary, a stage top incubator maintaining 37°C or 28°C and 5% CO2 was used.
Live Fluorescence imaging of EB3-GFP in 2D and GFP-Rab6A in 3D was performed on the same spinning disk confocal configuration. Acquisitions were performed at five frames/s during 2 min.
Alternatively, 2D samples imaging was performed using widefield fluorescence illumination on a Nikon Eclipse 80i upright microscope equipped with a CoolSNAP HQ2 CCD camera (Photometrics), an Intensilight C-HGFI precentered fiber illuminator (Nikon), a Plan Apo VC 100x NA 1.4 oil or 60x NA 1.4 oil and driven by Nikon NIS Br software.
Live-cell TIRF imaging was performed on a Nikon Eclipse Ti-E inverted microscope equipped with perfect focus system (Nikon), a CFI Apo TIRF 100X oil objective (Nikon), a TI-TIRF-E motorized TIRF illuminator (Nikon), a QuantEM 512SC EMCCD camera (Photometrics, Roper Scientific) and a stage top incubator maintaining 37°C and 5% CO2 (Tokai hit). The system was controlled with MetaMorph 7.5 software (Molecular Devices, San Jose, CA).
Gated STED imaging was performed with Leica TCS SP8 STED 3X microscope driven by LAS X controlling software and using HC PL APO 100x/1.4 oil STED WHITE objective, 633 nm white laser for excitation and 775 nm pulsed lased for depletion. Images were acquired in 2D STED mode with vortex phase mask. Depletion laser power was equal to 90% of maximum power and an internal Leica GaAsP HyD hybrid detector with a time gate of 1 ≤ tg ≤ 8 ns was used.
FRET analysis of RhoA and Rac1 biosensorsRequest a detailed protocol
The activities of RhoA and Rac1 were measured using previously described Rho single chain biosensor (Fritz et al., 2013) and Rac1 single chain biosensor (Moshfegh et al., 2014) using ratiometric FRET between mTFP1/mCerulean and mVenus. Live FRET imaging was performed on Leica TCS SP8 microscope equipped with spectral detection using HC PL APO 100x/1.4 oil STED WHITE objective. 440 nm pulsed laser (40MHz) was used for the excitation. Two channels were acquired simultaneously using hybrid detectors in the spectral ranges of donor 450–500 nm and acceptor 515–550 nm. Images were acquired with a scanning velocity of 100 Hz and eight line average scans, a pixel size of 0,416 μm and dimensions of 256 × 256 pixels. Donor and acceptor images were convolved with a Gaussian of 1 pixel, background subtracted and FRET-index image was calculated using ImageJ macro according to the formula: ,
where IA and ID correspond to the pixel intensity values of acceptor and donor images. Cell outlines were determined from the thresholded acceptor image and the average FRET-index value was calculated per cell.
Image preparation and analysisRequest a detailed protocol
For image preparation, we used ImageJ for adjustments of levels and contrast, maximum intensity projections, stitching (with pairwise stitching plugin) and thresholding to create binary mask used for circularity measurements and particle detection.
Kymographs of MT plus end and Rab6 3D dynamics were made using the KymoResliceWide plugin of ImageJ software and analyzed using the same software. For MTs, only length changes ≥ 0.3 μm between two consecutive time points were considered as growth or shortening events, while changes < 0.3 μm were considered as a pause event; only the events starting and finishing within the recording were analyzed. Velocity was calculated for each growth event and then averaged. Catastrophe frequency was calculated by dividing the number of catastrophes (transition from growth or pause to shortening) by the sum of growth and pause durations.
EB3 comets and CAMSAP2 stretches were automatically detected on thresholded pictures using the Particle Analysis plugin of ImageJ software. Their number (EB3) and surface (CAMSAP2) were quantified and reported to the cell surface area.
EB3 enrichment at the Golgi (/centrosome) was calculated as the ratio between the average EB3 intensity in a 2 µm diameter-circle drawn around Golgi mini-stacks (/centrosome) and the average EB3 intensity in the cytoplasm.
Golgi dispersion was calculated as in which SDi is the standard deviation of intensity and i the mean intensity.
For cell-cell junctions analysis (VE-Cadherin and ZO-1), ImageJ was used to plot intensity profiles along a manually drawn line across junctions. These profiles were then analyzed using the ‘area under curve’ function of GraphPad prism five and the maximum value of, as well as the area under the peaks were averaged.
F-actin staining in 3D was analyzed similarly by plotting intensity profiles along a 10 µm long rectangle drawn 5 µm away from the cell body using ImageJ and measuring the maximum intensity and the peak area using GraphPad prism 5. For stress fiber analysis, a customized ImageJ macro was used to trace the stress fibers and measure their length and width (Teeuw and Katrukha, 2015) (available at https://github.com/jalmar/CurveTracing).
All cell protrusions were manually traced with ImageJ software to quantify their number, length and spatial distribution. Polarity index was calculated as in which αi is the angle between the protrusioni and the longest protrusion and Li is the length of the protrusion. Cell masks were analyzed using the Particle Analysis plugin of ImageJ software to measure cell circularity, calculated as in which A and P are the area and the perimeter of the cell mask, respectively.
Time-lapse imaging of monolayer wound healing assays was analyzed using Manual Tracking and Chemotaxis Tool plugin of ImageJ software to measure the velocity and directionality (the ratio between the Euclidian and accumulated distance) of cell movement.
Time-lapse imaging of Rab6 and KIF13B was analyzed using ImageJ software with a recently developed plugin (Yao et al., 2017). The SOS plugin combined two procedures: particle detection and particle linking: SOS detector 3D module as detector and SOS linker (NGMA) module as linker were used. The tracking results were then processed using MTJ (MTrackJ) Simple Track Segment module to remove the non-directional tracks and to analyse speed, duration and length of the runs. MTJ Measure Region was used to determine the number of tracks contained in the cell front area, defined as the 45° sector facing the cell lamellipodia and originating from the center of the nucleus. Percentage of tracks in the front was calculated by dividing this value by the total number of tracks and correcting for area differences.
Radial representations of time-lapse images of protrusion formation in 3D were made using successively the Radial Reslice, Reslice and Minimum Intensity projection functions of ImageJ software.
ImageJ Radial Profile plugin was used to measure the distribution of CAMSAP2 and GM130 signal intensity along the radius in a 20 µm radius-circle originating from the Golgi center and each profile was normalized as .
For Golgi enrichment index of α- or γ-tubulin, z-maximum projection of α-, γ-tubulin or GM130 channel was thresholded using ImageJ to create a binary mask to delineate the cell or the Golgi area. The difference between the average intensity within the Golgi area and in the area outside the Golgi was divided by the intensity outside the Golgi and expressed in percent.
α-tubulin or CAMSAP2 enrichment in the longest protrusion was calculated as the ratio between the fluorescence intensity in a 0.75 µm diameter-circle drawn 8 µm away from the cell body or in the manually drawn area in the longest protrusion and in the other protrusions, within z-maximum projections and averaged per cell.
Time-lapse imaging of zebrafish venous sprouting was analyzed by manually drawing the vector corresponding to a sprout for each time point and measuring sprout length and angle using the measure function of ImageJ software. When needed, 3D color-coded stacks were used to more easily isolate the venous sprout. When the geometry of a sprout did not fit a straight line, a segmented line was used for length measurement and the straight line between sprout extremities for the angle determination. Growth persistence of a sprout elongation event corresponds to the mean value of length variations (Δ Length) between two consecutive time frames (every 10 min), whereas directional persistence was calculated for each frame as , where Δα represents the angle variation between two consecutive time frames, and then averaged per growing event.
To analyze MT radiality, images of fluorescently labeled MTs were separated into radial and non-radial components using customized ImageJ macro (Katrukha, 2017). (available at https://github.com/ekatrukha/radialitymap; copy archived at https://github.com/elifesciences-publications/radialitymap). First, local orientation angle map was calculated for each pixel using OrientationJ plugin (Püspöki et al., 2016). We used ‘cubic spline gradient’ method and tensor sigma parameter of 6 pixels (0.4 µm). The new origin of coordinates was specified by selecting the centrosome position in a corresponding channel, or the brightest spot in case of centrinone treatment. Radial local orientation angle was calculated as a difference between the local orientation angle and the angle of vector drawn from the new origin of coordinates to the current pixel position. A radial map image was calculated then as an absolute value of the cosine of the radial local orientation angle at each pixel providing values between zero and one. A non-radial map image was calculated as one minus radial map. Both maps were multiplied with the original image to account for different signal intensities; the two maps illustrate separated radial and non-radial image components.
The radial profile of the signal in the non-radial map image (normalized to the maximum signal of the original picture) was built using ImageJ and used to calculate the average non-radial proportion of the MT network. To avoid the artifacts of the cell center (very high signal) and border (MT bending), only a circular section around the reference point was used in the averaging (from 2.5 µm to 15 µm).
All mentioned ImageJ plugins have source code available and are licensed under open-source GNU GPL v3 license.
Zebrafish experimentsRequest a detailed protocol
The Tg(fli1a:eGFP)y1 and Tg(Fli1ep:Lifeact-EGFP) (Phng et al., 2013) lines were raised according to EU regulations on laboratory animals. All animal experiments were approved by the animal welfare committee of the University of Liege (protocol number 14–1556, laboratory agreement number LA 1610002). Knockdown experiments were performed by injecting embryos at the one- to two-cell stage with 6 ng of Camsap2b morpholino. The following Camsap2b-splice blocking morpholino sequence was used: ATACAGATGgcaagtcttttacatc.
For rescue experiment, a Morpholino insensitive human Camsap2 was built by overlapping PCR-based strategy (ATACAGATG transformed into ATTCAAATG), inserted into PSC2 + vector, linearized, in vitro transcribed and injected at 50 ng/µl.
For RT-PCR, cDNA was generated from total RNA extracted from zebrafish embryos with Trizol reagent (Thermo Fisher Scientific) using RevertAid RT Kit (Thermo Fisher Scientific) with random hexamer primers. After DNAse treatment (Thermo Fisher Scientific), cDNA was submitted PCR amplification followed by gel electrophoresis analysis using the following primers: ACTTCAGCAGGGCCAAGATA and TGTCACAGCCTCTTCAGCAT.
Alternatively, cDNA was submitted to quantitative real-time (q)PCR using Sybrgreen technology (Applied Biosystems, Foster City, CA) on a ViiA7 apparatus (Applied Biosystems). ELFA was used as reference gene to quantify the relative expression of the exon2 of Camsap2b using the ΔΔCt method with three alternative primers pairs. Primer sequences were as followed
For analyses of vascular structure formation, screening was performed under a fluorescence stereomicroscope whereas confocal pictures and movies were performed on artificially dechorionated embryos between 30 and 48hpf embedded in low melting point agarose (0.8%).
Statistical analyses were performed using GraphPad Prism five or Excel and significance was assessed using Mann-Whitney U-, Chi square with Yates correction- and Student’s t- two-tailed paired and unpaired tests. The statistical test used as well as the sample size is indicated in the figure legends. All data are shown using box plots where rectangles represent the second and third quartiles, contain a line corresponding to the median value and are extended with whiskers showing the minimum and maximum, except in Figure 6—figure supplement 1B and Figure 9D, which depict mean ± SEM. In Figure 1—figure supplement 1B and Figure 4—figure supplement 1B, the mean value ± SEM is indicated within the pictures. No explicit power analysis was used to determine sample size and no masking was used for analysis.
Microtubule minus-end-targeting proteinsCurrent Biology 25:R162–R171.https://doi.org/10.1016/j.cub.2014.12.027
Centrosome nucleates numerous ephemeral microtubules and only few of them participate in the radial arrayCell Biology International 39:1203–1216.https://doi.org/10.1002/cbin.10492
Role of the cytoskeleton in formation and maintenance of angiogenic sproutsJournal of Vascular Research 48:369–385.https://doi.org/10.1159/000324751
ERK regulates Golgi and centrosome orientation towards the leading edge through GRASP65The Journal of Cell Biology 182:837–843.https://doi.org/10.1083/jcb.200805045
The centrosome in cells and organismsScience 335:422–426.https://doi.org/10.1126/science.1209037
Rac1 and Aurora A regulate MCAK to polarize microtubule growth in migrating endothelial cellsThe Journal of Cell Biology 206:97–112.https://doi.org/10.1083/jcb.201401063
One-dimensional topography underlies three-dimensional fibrillar cell migrationThe Journal of Cell Biology 184:481–490.https://doi.org/10.1083/jcb.200810041
A role for cytoplasmic dynein and LIS1 in directed cell movementThe Journal of Cell Biology 163:1205–1211.https://doi.org/10.1083/jcb.200310097
Vascular morphogenesis in the zebrafish embryoDevelopmental Biology 341:56–65.https://doi.org/10.1016/j.ydbio.2009.10.035
Centrosome positioning in polarized cells: common themes and variationsExperimental Cell Research 328:240–248.https://doi.org/10.1016/j.yexcr.2014.09.004
Microtubules in cell migrationAnnual Review of Cell and Developmental Biology 29:471–499.https://doi.org/10.1146/annurev-cellbio-101011-155711
Coordinating cell behaviour during blood vessel formationDevelopment 138:4569–4583.https://doi.org/10.1242/dev.062323
Distribution of microtubule organizing centers in migrating sheets of endothelial cellsThe Journal of Cell Biology 91:589–594.https://doi.org/10.1083/jcb.91.2.589
Disconnecting the Golgi ribbon from the centrosome prevents directional cell migration and ciliogenesisThe Journal of Cell Biology 193:917–933.https://doi.org/10.1083/jcb.201011014
Angiogenic network formation in the developing vertebrate trunkDevelopment 130:5281–5290.https://doi.org/10.1242/dev.00733
Laser irradiation of centrosomes in newt eosinophils: evidence of centriole role in motilityThe Journal of Cell Biology 98:1999–2010.https://doi.org/10.1083/jcb.98.6.1999
Centrosomes, microtubules and neuronal developmentMolecular and Cellular Neuroscience 48:349–358.https://doi.org/10.1016/j.mcn.2011.05.004
Excess centrosomes disrupt endothelial cell migration via centrosome scatteringThe Journal of Cell Biology 206:257–272.https://doi.org/10.1083/jcb.201311013
Excess centrosomes perturb dynamic endothelial cell repolarization during blood vessel formationMolecular Biology of the Cell 27:1911–1920.https://doi.org/10.1091/mbc.E15-09-0645
Dynein drives nuclear rotation during forward progression of motile fibroblastsJournal of Cell Science 121:3187–3195.https://doi.org/10.1242/jcs.033878
Orientation and function of the nuclear-centrosomal axis during cell migrationCurrent Opinion in Cell Biology 23:579–588.https://doi.org/10.1016/j.ceb.2011.08.001
Ninein is expressed in the cytoplasm of angiogenic tip-cells and regulates tubular morphogenesis of endothelial cellsArteriosclerosis, Thrombosis, and Vascular Biology 28:2123–2130.https://doi.org/10.1161/ATVBAHA.108.169128
A Trio-Rac1-Pak1 signalling axis drives invadopodia disassemblyNature Cell Biology 16:571–583.https://doi.org/10.1038/ncb2972
Ninein is released from the centrosome and moves bi-directionally along microtubulesJournal of Cell Science 120:3064–3074.https://doi.org/10.1242/jcs.010322
Distinct ECM mechanosensing pathways regulate microtubule dynamics to control endothelial cell branching morphogenesisThe Journal of Cell Biology 192:321–334.https://doi.org/10.1083/jcb.201006009
Spheroid-based in vitro angiogenesis modelMethods in Molecular Biology 1430:167–177.https://doi.org/10.1007/978-1-4939-3628-1_11
Filopodia are dispensable for endothelial tip cell guidanceDevelopment 140:4031–4040.https://doi.org/10.1242/dev.097352
Transforms and Operators for Directional Bioimage Analysis: A SurveyAdvances in Anatomy, Embryology, and Cell Biology 219:69–93.https://doi.org/10.1007/978-3-319-28549-8_3
The centrosome-Golgi apparatus nexusPhilosophical Transactions of the Royal Society B: Biological Sciences 369:20130462.https://doi.org/10.1098/rstb.2013.0462
Control of centriole length by CPAP and CP110Current Biology 19:1005–1011.https://doi.org/10.1016/j.cub.2009.05.016
Migrating fibroblasts perform polarized, microtubule-dependent exocytosis towards the leading edgeJournal of Cell Science 116:4513–4519.https://doi.org/10.1242/jcs.00748
Microtubule-induced cortical cell polarityGenes & Development 21:483–496.https://doi.org/10.1101/gad.1511207
Visualization of microtubule growth in cultured neurons via the use of EB3-GFP (end-binding protein 3-green fluorescent protein)The Journal of Neuroscience : The Official Journal of the Society for Neuroscience 23:2655–2664.
CPAP is a cell-cycle regulated protein that controls centriole lengthNature Cell Biology 11:825–831.https://doi.org/10.1038/ncb1889
Centrosome positioning in vertebrate developmentJournal of Cell Science 125:4951–4961.https://doi.org/10.1242/jcs.038083
A newly identified myomegalin isoform functions in Golgi microtubule organization and ER-Golgi transportJournal of Cell Science 127:4904–4917.https://doi.org/10.1242/jcs.155408
Golgi positioningCold Spring Harbor Perspectives in Biology 3:a005322.https://doi.org/10.1101/cshperspect.a005322
A primary role for Golgi positioning in directed secretion, cell polarity, and wound healingMolecular Biology of the Cell 20:1728–1736.https://doi.org/10.1091/mbc.E08-10-1077
KIF13B regulates angiogenesis through Golgi to plasma membrane trafficking of VEGFR2Journal of Cell Science 127:4518–4530.https://doi.org/10.1242/jcs.156109
Automated Analysis of Intracellular Dynamic ProcessesMethods in Molecular Biology 1563:209–228.https://doi.org/10.1007/978-1-4939-6810-7_14
Centrosome reorientation in wound-edge cells is cell type specificMolecular Biology of the Cell 13:1871–1880.https://doi.org/10.1091/mbc.01-11-0539
Golgi as an MTOC: making microtubules for its own goodHistochemistry and Cell Biology 140:361–367.https://doi.org/10.1007/s00418-013-1119-4
Pekka LappalainenReviewing Editor; University of Helsinki, Finland
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
Thank you for submitting your article "Control of endothelial cell polarity and sprouting angiogenesis by non-centrosomal microtubules" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Fiona Watt as the Senior Editor and Pekka Lappalainen as the Reviewing Editor. The reviewers have opted to remain anonymous.
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
This manuscript focuses on the roles of different types of microtubule anchoring/stabilization mechanisms (centrosomally anchored versus Golgi-anchored CAMSAP2-stabilized versus CAMPSAP2-stabilized not Golgi-anchored) in the polarization and movement of endothelial cells in 2D and 3D environments. The major finding is that, especially in soft 3D environments, centrosomal microtubules are not essential for establishment of polarity, whereas CAMSAP2-stabilized, non-centrosomal microtubules are required for persistent elongation of the cell body. Moreover, the authors show that CAMSAP2 plays a role in sprouting angiogenesis in vivo.
These findings are interesting, and the analysis and quantification methods are well designed and state of the art. The topic of the interplay between centrosomal, Golgi-anchored, and non-centrosomal microtubules and the role of the CAMSAPs in the dynamics of non-centrosomal arrays is also of high current interest. However, there are few points that should be addressed to further strengthen the study.
1) The authors should examine at which level the cell morphology is affected in the absence of CAMSAPs (i.e. do CAMSAP-dependent microtubules generate and stabilize protrusions in ECs or do they stabilize protrusions indirectly by regulating the actin cytoskeleton). The authors should thus more thoroughly examine the organization and dynamics of the actin cytoskeleton in control vs. CAMSAP knockdown cells. Because CAMSAP knockdown cells appear to have more stress fibers compared to control cells (based on the images presented in Figures 4B and C), the authors could also use e.g. blebbistatin to examine whether the differences in the number and persistence of protrusions between control and CAMSAP knockdown cells could result from their different contractility.
2) Unlike stated in the third paragraph of the Discussion, the authors cannot conclude from their data that CAMSAP2 depletion completely blocks sprouting from spheroids. There are plenty of short sprouts (in line with protrusive activity of isolated ECs, which can make protrusions), but not very long, single ones.https://doi.org/10.7554/eLife.33864.052
1) The authors should examine at which level the cell morphology is affected in the absence of CAMSAPs (i.e. do CAMSAP-dependent microtubules generate and stabilize protrusions in ECs or do they stabilize protrusions indirectly by regulating the actin cytoskeleton). The authors should thus more thoroughly examine the organization and dynamics of the actin cytoskeleton in control vs. CAMSAP knockdown cells. Because CAMSAP knockdown cells appear to have more stress fibers compared to control cells (based on the images presented in Figures 4B and C), the authors could also use e.g. blebbistatin to examine whether the differences in the number and persistence of protrusions between control and CAMSAP knockdown cells could result from their different contractility.
It is true that there is a tight interplay between the microtubule and the acto-myosin cytoskeleton that regulates cell shape and motility. In particular, the effect of microtubule signaling on Rho GTPase-mediated control of cell protrusion and contractility is often thought to represent the most important microtubule function during cell migration.
In order to find out whether the role of CAMSAP2-stabilised non-centrosomal microtubules was mediated by the regulation of the acto-myosin cytoskeleton, we performed a new set of 2D and 3D experiments to:
- assess the impact of CAMSAP2 depletion on the actin cytoskeleton
- examine whether a potential change in contractility could explain CAMSAP2-associated defects.
The entire new Figure 4 (complemented with two supplement figures) now addresses this question in detail. We found that whereas CAMSAP2-knockdown cells exhibited a small increase in the abundance of stress fibers in 2D (Figure 4A, Figure 4—figure supplement 1A), this change was not associated with alterations in Rho and Rac1 activation pattern, the arrangement of cell-cell junctions or formation of myosin IIb-enriched retracting rear (Figure 4B, Figure 4—figure supplement 1B, C, already included the first version of the manuscript). More importantly, when looking in the context of a 3D environment, we now found that within the protrusions, the actin cytoskeleton that is mainly cortical was unaltered following CAMSAP2 depletion (Figure 4C, Figure 4—figure supplement 1D).
In agreement with previous 3D observations showing that cortical contractility counteracts protrusion formation (Sanz-Moreno et al., 2008, Fischer et al., 2009, Bouchet et al., 2016), we observed longer and more persistent protrusions, as well as more and longer spheroid sprouts after treatment with the ROCK inhibitor Y-27632 and the myosin II inhibitor blebbistatin (Figure 4E, F, Figure 4—figure supplement 2A-F). Nevertheless, even in a background with reduced contractility, CAMSAP2 silencing still severely affected protrusion persistence, cell elongation and sprout formation (Figure 4D, E, F, Figure 4—figure supplement 2A-F). Taken together, these data show that CAMSAP2-stabilised microtubules are controlling protrusion persistence and organization without significantly affecting acto-myosin cytoskeleton and contractility, and we added this consideration to the Discussion (second paragraph).
2) Unlike stated in the third paragraph of the Discussion, the authors cannot conclude from their data that CAMSAP2 depletion completely blocks sprouting from spheroids. There are plenty of short sprouts (in line with protrusive activity of isolated ECs, which can make protrusions), but not very long, single ones.
We agree that CAMSAP2-depleted spheroids could still form short sprouts and that saying that CAMSAP2 silencing almost completely blocked sprouting is a too strong statement. In total agreement with the reviewers’ comment, we think that the “protrusive activity” of CAMSAP2depleted ECs, important for protrusion initiation, is not affected, explaining the presence of short sprouts. Nevertheless, whereas the lack of protrusion persistence in isolated ECs “only” impaired their organization, the inability of CAMSAP2-silenced spheroids to mature larger sprouts caused their regression (Figure 3—figure supplement 1A). This ultimately led to a difference in the severity of the phenotypes that we do find striking (for example, if we compare Figure 2G with Figure 3—figure supplement 1C).
We rephrased our statement to more accurately reflect these considerations (subsection “CAMSAP2 is required for maintaining non-centrosomal MTs and cell migration in ECs”, last paragraph; subsection “CAMSAP2 is required for stabilization of one major cell protrusion”, first paragraph; subsection “Centrosome removal promotes cell polarization in the absence of CAMSAP2”, last paragraph; Discussion, second and third paragraphs).https://doi.org/10.7554/eLife.33864.053
Article and author information
European Research Council (Synergy 609822)
- Anna Akhmanova
Nederlandse Organisatie voor Wetenschappelijk Onderzoek (ALW Open Program grant 824.15.017)
- Anna Akhmanova
H2020 Marie Skłodowska-Curie Actions (IEF fellowship)
- Maud Martin
China Scholarship Council (PhD fellowship)
- Jingchao Wu
Fonds De La Recherche Scientifique - FNRS (FRIA fellowship)
- Alexandra Veloso
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
Animal experimentation: All animal experiments were approved by the animal welfare committee of the University of Liege (protocol number 14-1556, laboratory agreement number LA 1610002).
- Pekka Lappalainen, University of Helsinki, Finland
- Received: November 26, 2017
- Accepted: March 13, 2018
- Accepted Manuscript published: March 16, 2018 (version 1)
- Version of Record published: April 13, 2018 (version 2)
© 2018, Martin et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
- Page views
Article citation count generated by polling the highest count across the following sources: Crossref, Scopus, PubMed Central.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
- Cell Biology
- Stem Cells and Regenerative Medicine
Following acute injury, the capillary vascular bed in the lung must be repaired to reestablish gas exchange with the external environment. Little is known about the transcriptional and signaling factors that drive pulmonary endothelial cell (EC) proliferation and subsequent regeneration of pulmonary capillaries, as well as their response to stress. Here, we show that the transcription factor Atf3 is essential for the regenerative response of the mouse pulmonary endothelium after influenza infection. Atf3 expression defines a subpopulation of capillary ECs enriched in genes involved in endothelial development, differentiation, and migration. During lung alveolar regeneration, this EC population expands and increases the expression of genes involved in angiogenesis, blood vessel development, and cellular response to stress. Importantly, endothelial cell-specific loss of Atf3 results in defective alveolar regeneration, in part through increased apoptosis and decreased proliferation in the endothelium. This leads to the general loss of alveolar endothelium and persistent morphological changes to the alveolar niche, including an emphysema-like phenotype with enlarged alveolar airspaces lined with regions that lack vascular investment. Taken together, these data implicate Atf3 as an essential component of the vascular response to acute lung injury that is required for successful lung alveolar regeneration.
- Cell Biology
- Microbiology and Infectious Disease
African trypanosomes proliferate as bloodstream forms (BSFs) and procyclic forms in the mammal and tsetse fly midgut, respectively. This allows them to colonise the host environment upon infection and ensure life cycle progression. Yet, understanding of the mechanisms that regulate and drive the cell replication cycle of these forms is limited. Using single-cell transcriptomics on unsynchronised cell populations, we have obtained high resolution cell cycle regulated (CCR) transcriptomes of both procyclic and slender BSF Trypanosoma brucei without prior cell sorting or synchronisation. Additionally, we describe an efficient freeze–thawing protocol that allows single-cell transcriptomic analysis of cryopreserved T. brucei. Computational reconstruction of the cell cycle using periodic pseudotime inference allowed the dynamic expression patterns of cycling genes to be profiled for both life cycle forms. Comparative analyses identify a core cycling transcriptome highly conserved between forms, as well as several genes where transcript levels dynamics are form specific. Comparing transcript expression patterns with protein abundance revealed that the majority of genes with periodic cycling transcript and protein levels exhibit a relative delay between peak transcript and protein expression. This work reveals novel detail of the CCR transcriptomes of both forms, which are available for further interrogation via an interactive webtool.